MISR/NDAI image processing project abstract
-------------------------------------------

This project consists of developing and applying software to estimate the
surface roughness (in RMS (Root Mean Squared) deviation from a local plane) of
the Earth, especially in those regions which are covered by ice and snow for
most of the year, by the use of remote sensing data from satellite images.  As a
test case, satellite image data acquired by the MISR (Multi-angle Imaging
Spectro Radiometer, http://www-misr.jpl.nasa.gov/) instrument from cameras using
3 different near-simultaneous angles of view as the spacecraft (Terra/NASA) made
repeated orbits over Greenland during the summers of 2002 and 2007 were used.
Composite images were made of Greenland from thousands of images, and from
combinations of images used to calculate the NDAI (Normalized Difference Angular
Index) which shows the relative ratio of back-scattered to forward-scattered
light from the surface.  As a reference, ATM (Airborne Topographic Mapping)
LIDAR (LIght Detection And Ranging) data were obtained for several localized and
distributed areas over Greenland which determined the RMS roughness to an order
of centimeters.  The ATM RMS and satellite NDAI and other image data were
correlated in 2-4 dimensions to yield a quantitative automatic prediction model
for estimating the RMS values at most locations covered by the satellite data
alone, and well away from the ATM test sites. One interesting feature of this
prediction is that it accurately estimates roughness of the surface to within
just a few centimeters by using image data which have a spatial resolution of
several hundred meters.  Further work involves applying the technique to other
snow and ice surfaces, and in automatically tracking features which have little
image contrast (e.g. glaciated fronts, ablation, accumulation, and transition
zones) by virtue of their calculated 3d surface properties.